AI and Knowledge Management with Derek Crager

In this episode of Definitely Maybe Agile, Peter Maddison and David Sharrock are joined by Derek Crager, a seasoned engineer turned AI entrepreneur who shares his journey from blue-collar work to building AI-powered knowledge management solutions. Derek discusses how AI is transforming workplace onboarding, knowledge transfer, and personal productivity, drawing parallels between today's AI revolution and the early days of the internet.

Derek brings a practical perspective on implementing AI in enterprise environments, focusing on his company's voice-powered AI assistant "Pocket Mentor" that helps organizations capture tribal knowledge and streamline employee onboarding. The conversation explores the challenges of extracting expertise from subject matter experts, the importance of having clear business outcomes when adopting AI, and advice for students navigating career choices in an AI-driven world.

Key Takeaways:

• Focus on Business Outcomes Over Buzzwords - Don't implement AI just to check a box or follow trends. Instead, identify specific business problems (like inefficient onboarding or knowledge retention) and then explore how AI can provide practical solutions to those challenges.
• AI Excels at Knowledge Augmentation and Accessibility - AI's greatest strength lies in making organizational knowledge instantly accessible 24/7, without judgment, and at a consistent quality level. This is particularly valuable for onboarding new employees and capturing tribal knowledge from subject matter experts before it walks out the door.
• Take Time to Learn and Evaluate Before Jumping In - The AI wave mirrors the early internet adoption curve and will continue evolving over the next 10-15 years. Organizations and individuals have time to thoughtfully evaluate AI solutions rather than rushing to adopt the first available option, emphasizing the importance of learning practical applications before implementation.

Chapters

• 0:04 Episode Introduction
• 1:13 Meeting Derek Krager
• 3:45 Technology Evolution and AI's Current State
• 7:16 Education, AI, and Future Career Paths
• 12:02 AI for Individual and Team Productivity
• 18:22 Onboarding Solutions and Voice AI Interface
• 24:12 Extracting Tribal Knowledge
• 26:42 Final Thoughts and Key Takeaways

Transcript

Peter (0:04): Welcome to Definitely Maybe Agile, the podcast where Peter Maddison and David Sharrock discuss the complexities of adopting new ways of working at scale. Hello everybody, so welcome once again, and today I'm joined by my good friend Dave, as always, and we're joined today by Derek. So Derek, why don't you introduce yourself to our audience?

Derek (0:24): Yeah, certainly. Hi Peter, Hi Dave, I'm Derek Crager. I'm 58 as of yesterday. I spent time after high school in blue-collar work to transition to engineering and process engineer and then on into the learning side. I tell people I used to get my hands dirty for a living and today I teach people how to get their hands dirty for a living. But I've jumped into the world of AI because it's headlines and who isn't in AI these days?

Peter (0:55): Yeah, it's the way to sell every project out there.

Dave (0:59): Certainly the introduction, isn't it? So did you want to maybe... Derek, just what drew you into AI from where you were at? I mean, yes, it's on everybody's mind over the last few years, but I think you've been involved a little bit longer than that, right?

Derek (1:13): My background in technology actually started back in the 80s. I think I was writing programs on a Texas Instrument TI-99/4A, back when the program you'd save it to an audio cassette tape and then I would, I remember mailing it. You put a stamp on an envelope and you put it in the box at the end of your driveway and it went to Texas. And for every program I sent in I got five back. But that's probably the core.

But the last technology wave that came through, Dave, is the Internet—the world of information, the World Wide Web. It just seems like just a tick above what we're already doing. So I got involved in the World Wide Web and I'm told I had one of the first 100 online stores ever. But honestly there's no way to check it back then, let alone today. But I had sold, used to sell, some Super Bowl shirts and hats and shorts back from 1990 to 1993. And I'd sold on five different continents. And it came to a point where my distributor said "Hey, we found this internet thing, so we're going to sell. We don't want you to."

Derek (3:00): Long story short, I said buy out my inventory because I thought I had saturated the market. Swear to God, I was so dense and so ignorant back then and it really wasn't until 2006, when the mobile phones came out, that everybody had access to the internet. So, even though it seemed like I got out early, I really got out way too early.

So you asked me what got me into AI? It's the pursuit of technology and the fact that today's wave is an AI wave. It mirrors that wave we had back in 1990. So to me it feels like it's 1990 or 1991 right now, but I have better eyesight, so I can see ahead. And even though AI is all a buzzword, it's really only the buzzword for those on the cutting edge and those that are looking to invest in the next big thing. So I really don't even see AI peaking for another 10, 15 years. I just wanted to be there.

Peter (3:45): What sort of things are you looking at applying it to?

Derek (3:45): Great question. As a personal support—we've had these machines, whether it be a speaker on the desk from Amazon, or a Siri on our Apple phone, or a Google on our Android phone, or fill in the blank, Samsung has theirs. They're trying to get into play, but I think it's going to be a great personal assistant. It's somebody... I think it's a captive audience, to be honest. It's somebody to talk to 24/7, that's going to answer the phone every time we call. So, from a personal standpoint, I think it's an empowering technology, I think it augments our knowledge base through just instant access, and I think everything's going to be leveraged around that.

Peter (4:22): It's kind of interesting. We interact with AI-driven machines quite differently to the way we interact with each other. There's some of the pieces from people I've been talking to in the industry, like when we're more willing to interrupt a machine, so we're more willing to hit stop. "No, that's not what I meant. You're giving me a bunch of garbage I don't want to listen to," and so you'll hit stop, which we're less likely to do to another human. We're more considerate of wanting to listen to what they're going to say, right?

Derek (4:48): Well, I agree, I agree. I would maybe flip a coin on the consideration part, because from humans it comes from being judged, and there's actually a New England Journal of Medicine has a paper on this that says that people get better response and get better benefits out of an AI-based counselor/psychiatrist/counseling method because they don't feel judged and they can get to the point much sooner of just spilling the tea and sharing what's on their mind, whereas when they interact in a professional sense, first there's a barrier of even getting to a human professional, of time and cost and access. But once they get to that professional, usually they said, it's like five or six one-on-ones before they can actually even expose themselves, get it off their chest. So yeah, I agree with you, Peter, it's just it's more accessible and that really helps.

Dave (5:50): You've described... I mean, maybe it's just I've been scrolling in the wrong direction, but I'm seeing a lot of news about some of the downsides of AI and one of the ways that that often ends up getting discussed is normally somebody in the conversation has kids just either going into university, at university or leaving university, and the question comes around what on earth should those kids be studying? What should people be studying at university at degree level, given, as you see where things are going with AI over the next 10 to 15 years, should we be changing the majors that we're looking at right now as considering the high value majors, or should we just carry on doing what we've been doing over the last 5, 10, 15 years?

Derek (6:36): Well, that's a huge question and I can only speak from my perspective of my experience. First, I just got a reference that everything that I've been taught, whatever someone studies in university, within five years 40% at most are even employed and have a career that had anything related to that topic. So I don't think it's an "I'm putting all my eggs in one basket" decision. I think it's one of those "I'm going in and I'm going to learn something, so just learn what can be leveraged elsewhere."

My daughter, for example, she went into psychology and the joke is there, with any undergrad degree you can get a minimum wage job at McDonald's type of thing. But she followed up into human factors and she ended up as a psychologist getting hired at a technology company. So it's just weird how these connections get made.

Derek (7:34): Back when I was at university, I looked at, I wanted to learn how to program, but the code, the program languages that were available were 30 and 40 and 50 years old and they weren't even applicable. So I think that today's day and age, that "don't put all your eggs in one basket" is key. There's so much more that can be learned on the front, on the cutting edge, in Discord, in Reddit, in all the free training that Google offers and IBM offers and the universities that offer all this free training. I would say augment what you're learning at university with that, because that will give you the reality of where we're at today and that can help you to find your own North Star on what you should be studying at university.

Peter (8:26): I think that's all really great advice. I think I would agree that there's the, with all the free resources out there too, all the things that you can draw upon to help expand that knowledge beyond even what school is teaching. And I certainly am not doing what I studied in university. Nowhere close. I barely remember most of it, to be honest.

Derek (8:46): What did you study, Peter?

Peter (8:48): I did dual honors maths and physics and I've got a master's in microwave solid state...

Derek (8:46): Okay, you're putting that to use right now.

Peter (9:06): Yeah, totally, totally. We are naturally because there are lots of... that's fantastic but not in my day-to-day job.

So it's interesting as we think of that too. The AI brings a lot of new capabilities to the table, which allows us to very quickly prototype and bring new solutions to market, to be able to test and learn much, much faster. Something we talk about a lot on this show. Have you found that with the product you're building?

Derek (9:25): Well, iteration is fantastic. We can have the AI do the iteration for us. From an individual level, I iterate development, I iterate marketing, I iterate sales messages back and forth with AI. AI is a good memory keeper.

My entire life, because I'm on the spectrum, as they say, and diagnosed ADHD, I have searched for decades for a knowledge keeper, whether it be a trapper keeper back from the 80s—the papers in it—or the tablets as they develop, or paper and pen. I've never found anything that really works. But I'm able to use and leverage AI today just to organize myself. I can say, "Hey, I forgot something, remind me what that was." And it's that advisor tool that really helps. Being so close, I can soundboard.

Derek (10:26): In the old days, if I had to call up an SME on a subject, I'd have to find a time they're available and I'm available and plan ahead. But today we create our own SMEs and it's just instantly available. So, just like the internet wave wasn't so much about digitalization, it was about connecting us to knowledge instantly, kind of like the Akashic Records making it happen. And it's the iteration, the soundboarding between humans is really the power that the World Wide Web gave us. So AI at this next level, we're able to iterate one-on-one by ourselves, in addition to working with humans. So it allows me, in my case, to zero in and focus in on what I really want to say, or develop something the way I really want to develop it. Get that MVP out there and then, once I get it now, I can engage human contact and say "What do you think? Now let's talk about it." So I'm not wasting your time just on the developments.

Dave (11:30): Derek, I love the way that you're describing that sort of individual, focused, targeted support that... I mean we're all... everybody's working hard, we're all busy, we've all got lots of demands on our time and our attention and having that sort of voice or companion that can keep us focused makes everything a little bit easier to kind of maybe navigate through or keep on top of. What I wanted to kind of explore around that are maybe some examples you've seen with people who are using some of the products that your company has been putting out there, but specifically around how teams and/or individuals are using that. So you mentioned a little on the individuals. What are you seeing maybe around the teams side?

Derek (12:14): Yeah, thank you, Dave. For our product we actually focus on a voice AI interface, so the onboarding space is huge. So many companies have no onboarding solution. So one use case is to onboard employees. My wife mentioned to me just last week. She said she's worked at companies where they hire so many people in a day but they don't have enough onboarding buddies or managers or trainers to get them on board. So they end up literally their job is to go stand in the corner or go sit at the desk until somebody can get to them. They're getting paid.

Derek (12:54): So onboarding is a realm where it benefits the large companies like Amazon that I work for—I develop their onboarding training on the reliability, maintenance, engineering side—but also benefits the small companies. The large companies get economies of scale. The small companies get that "I can't dedicate a training team to train and we're going to have to do it one-off." So we have one iteration and by the time they use it on their third new hire, well, now half of it's not even applicable.

With AI interfacing this, we actually take those onboarding SOPs and those books and allow new hires to come in. It augments the training team. The training team still needs to identify opportunities and create those SOPs, but it's the AI interface to their knowledge base that allows them to bring in a onboarding... onboard a new employee and literally onboard them 80% faster because the onboarding employee can come in and they have in their ear somebody they can talk to. They have that onboarding buddy that doesn't cost $100,000 a year on somebody's payroll—pennies on the dollar type of thing—but it's leveraging all the information and it's the single point of truth. So instead of trying to update all the onboarding manuals that were printed last week or last year, it's digital and then the AI actually takes it and incorporates it and delivers it in the format needed for that individual. So 80% faster onboarding is definitely key in one area.

Dave (14:33): I'm just trying to translate it into a different perspective of reframing what you're describing there, Derek. So what I'm hearing you saying is it's almost as if it's an additional channel. I could either have a conversation with you, onboarding in your company and learning about that, maybe somebody else on your team or maybe a virtual training, or I can use an AI which is going to have a much more informal, interactive, conversational approach to the onboarding. Same material, same content, just at my pace, without being tied to other individuals stepping in and guiding through that process.

Derek (15:06): Oh, you nailed it, Dave, 100%, spot on. And here's something that the AI is that if I'm, as a human training, my 110th person for the day or the week, I'm just going through the motion and I don't care anymore. And the employee knows that I don't care anymore. But the thing is this anthropomorphization of AI through a conversation. It actually not only does it escape me being judged by asking stupid questions that we both touched upon earlier, but it actually cares.

Derek (15:40): Our AI is designed to be helpful, to check in to see if they're okay, to be patient and jovial even. It'll throw in a joke every now and then just to keep it lighthearted. And it delivers the same information, the highest quality information, consistently. So I know me as a human. I don't deliver my best deliverable at 100% every day. I would want to. I can try to, but when it goes to AI, it can deliver that best information at the right time and it does so without judgment. So yeah, from that individual's perspective, it's just like you shared there. It's available, it's accommodating, it's patient. It doesn't say "Hurry up and finish this or work on your own. I got to go clock out" because AI doesn't clock out, especially the voice interface Pocket Mentor product that we have.

Peter (16:39): It's always there with all the right information for any time, day or night.

Derek (16:43): Well, it's like the old accounting term G-I-G-O, right? Garbage in, garbage out. So it has to be set up properly and it's as easy to use as those Alexa, Siri, "Hey, Google" devices that are out there. But what we are, we're at the enterprise level and anybody that's going to use AI at the enterprise level—if once you dabble in it, you know it's not just about hooking up an LLM to your data. It goes beyond that, because you need to filter, you need to train, you need to set it up specifically, you have to put guardrails on it, you got to put containers around it so it accesses that information and only that information. So I think the use case could be "How do I speak to my phone or those Alexa products on the shelf?" But in this case, AI is to the point where we can focus that knowledge and the interfacing. So it's so lifelike that you're getting good information at the right time, and I think that's the enterprise difference today, something that we couldn't offer even a year ago.

Dave (17:50): Can I just follow up on that one? Do you have experience working with technology teams delivering some sort of digital product? Because what I'm hearing you describe becomes quite interesting when you look at developer environments. You look at organizations which are insisting that productivity gains are huge when you start using some of the AI tools around development. And yet certainly our experience is, yes, there are definitely engineers that kind of get a lift, but it's tool-based. It's not anybody's sort of speaking in your ear, if that's one side of it, but also it's almost like "We have to do this, otherwise we're going to get into lots of trouble because our productivity doesn't grow." And, Peter, I know you work in this space. You could probably speak a lot to this as well. But I'm just wondering what experience you've seen in technical development teams using products like Pocket Mentor.

Derek (18:41): Well, my background is I'm a project engineer, so I've been trained, got my PMP certification, have my agile certification. So the old school PMP style is let's map out the process and predict where we're going to be. The agile methodology which came about in the late 90s is well, let's start on something and we'll modify it as we get information in. So that's the basics of those two projects. Now, to answer your question, I can't say that I have a use case where I have, like a sprint team. Is that what you're thinking of on the development side?

Dave (19:22): Yes, exactly. A group of engineers, test engineers, whoever may be on that cross-functional team. But I just see there's... We already know we're working with teams that are already using some of the tools out there, but there's something quite intriguing about having basically a buddy along to run through different solutioning ideas or explore different solutions before you're actually building the solution that you're exploring.

Derek (19:50): Okay, Well, I know that there's AI tools outside of our product, because our product really is... It's called Pocket Mentor because it guides the individual. It's good for channeling the right information at the right time, but information that's already there. That's where we excel at. There are AI platforms that work with development teams, software development that allows creation of MVPs and processing and experimenting in code that accelerates that. I don't think that I personally have a solution for software developers. It falls outside the bounds of what we're at. I know that a lot of developers they have their own stack, that they work with their own preference and they're either already using what they want to use or they have no desire of changing how they're going to do it. So beyond that, I guess I'm not able to speak to the software development side. From my product's perspective, just knowing that AI is being leveraged, but it's more the hands-on interface, visually, versus the audible communication side of things.

Peter (21:08): It could be a whole other podcast, I think, on the various uses of AI in SDLC, the different elements where it comes in, the different things it could do, and the products that I'm involved in, which do look at software delivery teams and how they work together and give them advice on how to improve on a daily basis—this type of thing.

Peter (21:27): So, outside of the tooling and the creation of the code itself, it is a fascinating space and where we're seeing AI get brought in in lots of different ways, but there is varying different reports on how effective it is, and I think one of the key pieces seems to be are we actually getting the outcomes we expected to? So are we putting AI into place to actually get better business outcomes? Your example for the onboarding piece is a very key one, for example, where we know we've got this business problem. It's a problem for the business. We can use AI to help solve this problem, whereas we're also seeing where we're bringing in tooling that's AI powered but isn't actually necessarily targeted at solving a problem the business actually has. So there's other problems that we're not really solving by bringing in these pieces that, say, increase a certain aspect of the system but not the part that's actually needing to be increased.

Derek (22:24): Yeah, I agree on that 100%, Peter. I follow Sol Rashidi and she speaks about AI being brought in just to fill a gap. "Check it off. We're using AI for something." And she actually just posted this morning. She said there's three types of AI. If you're going to define AI, she calls it the AI trinity. She wanted to make sure that's trademarked her term, but her AI trinity talks about three parts of AI. One is automation, like doing things that are repetitive. And then the other one is augmentation, like augmenting knowledge of human workers and taking stress off the shoulders of that person, and that's where we come in. We're on the augmenting the knowledge part. And then the third part of that Trinity she mentioned is prediction—prediction or predictive analysis, taking that data and moving forward to that.

Peter (23:14): What do you think, Dave? We go from data to information, to knowledge. If it was that easy, right, are we not going back to that conversation about courses?

Dave (23:23): Again we go to university, we get knowledge, and it doesn't necessarily turn into anything that we can use.

Dave (23:30): I was going... I just wanted to like, as we're kind of spinning around a couple of these things... one of the again I keep coming at this from the organizations that we work with and one of the big headaches you'll often see is these subject matter experts and I know you've experienced this, Derek, in your work. Subject matter experts, these individuals that know everything about how to solve something and understanding how to get that tribal knowledge... again, a little bit like that data, information knowledge piece, how to take the information that's not written down, that's in somebody's head. They've spent whatever a decade or more working in an organization. How do you start addressing that problem of extracting that tribal knowledge so that the pocket mentor can pick that up and start providing that to other people? Where does that begin?

Derek (24:18): Well, that was a big gap because, even though we can take PDFs and manuals and operational SOPs and just ingest that because it's digital to digital. That was a big question, and what we design and how we handle that. Today, the SMEs that are out there, they're usually doing work of some kind. They could be developers, they could be hands-on machine repair, they could be office personnel. But generally those SMEs, their blind spot is that they know so much about the subject—hence the SME title, subject matter expert—that they can't explain it very well, and not without going through the motions.

So what we do, we have them call in our situation, they call a pocket mentor and they do an interview. If you want to see somebody's eyes light up, talk to an SME and ask them how to do something because, boom, their eyes get big and they smile with their entire face and they say, "Oh yeah, and you start here and you push this button, you do that, but don't forget about this and make sure that's in line and check that dashboard over there."

Derek (25:28): And then, when you get done, you just hit submit and they're just so excited. It's childlike, and I'm saying that in the most optimistic view of that. So we take that through the conversation and it's all recorded, on how to do something, and then at that point it just becomes digital information that we merge with the knowledge base. So now we're documenting that tribal knowledge, so not only do your SMEs have it, but everybody on your team has access to that same... We don't even call it tribal knowledge anymore. Now it's documented knowledge and it's available for everybody on the team 24/7. No sick days. And here's the big thing: Once all that knowledge is documented and saved, you don't have that knowledge walking out the door when somebody retires or transitions to another company.

Peter (26:22): Yeah, that makes so much sense. Okay, well, I think we're almost at time here. So, as we traditionally do on this show, I'm going to ask for each of us to provide what our favorite point or takeaway from this conversation is. So I'm going to start with you, Dave. So what's your favorite takeaway from this conversation?

Dave (26:42): So I'm going to go back to a question that I started... we started chatting to Derek about, which was what people should, what students should, study when they get to university, because I realized I never said anything. I don't have anything nearly as impressive as Peter's double maths physics major, but I ended up studying geophysics and earthquakes. And I can tell you categorically, absolutely I've not dealt with earthquakes in a number of decades, so that one was a real kind of interesting, I'd say realigning or resetting of that bar about what the challenge is for students going into universities today.

Peter (27:21): I think you are somewhat underselling your list of qualifications there, Dave, but anyway.

Dave (27:29): Unfortunately, I've still not worked with earthquakes for a couple of decades now.

Peter (27:34): That may be true. So, Derek, what point would you take?

Derek (27:38): Well, I was certain you're going to go to you next. I have another second or two to think about it, but you got me.

Peter (27:45): I get to go last.

Derek (27:49): He who has the gold has the power, all right. So in this case, it's the microphone. So I think my biggest takeaway is just about knowledge. I believe knowledge is power and knowledge is there for us.

Don't worry that AI, that you're going to miss the next big thing. It is a wave in the general sense, and the mirroring of how the AI wave is going to be accepted into our human realm is going to match, just like the internet wave did 35 years ago. It's starting with those that are early adopters and even though there's headlines that are saying it's going to solve our problems or it's going to cause all our problems, and they keep throwing around the word singularity as something we should all be afraid of or maybe embrace, who knows?

Derek (28:44): I think it's just important to know that, if you have, if you're adopting AI for your business or for yourself or for your children, as you're thinking ahead, if you're looking to adopt an AI strategy, just know don't take the first car that comes down the track. You have plenty of time. It's important to learn. Don't jump. It's not ready, fire, aim—you take it in steps and think about what you want to use AI for and then explore true practical scenarios where you can adopt that technology the way you want to adopt it and don't get blinded by all that. That's my takeaway.

Peter (29:25): I think that's a very good way of summing it up, I think. For my part, as you said, I get to go last, but the piece I would tell them is something you were just mentioning. There is the business outcome—what is it you're trying to do with all of this? So it's not enough to just take this and blindly apply it because it's the latest buzzword, the latest thing that everybody wants. It's really thinking about well, what can I do with this?

Peter (29:51): What can I do differently as a consequence of this, and starting to rethink the processes and systems that we have and how we interact with the different areas of our organization, rather than just saying, "Well, we're going to take whatever it is we did before and we're going to apply AI," because the guy above me is really asking me to work out how to do that, whereas it may not be necessarily the best thing to do. So, with that in mind, I'd like to say thank you very much, Derek. It's been a pleasure having you on the show and thank you, as always, Dave, and we'll wrap up there, and people can reach out to us at feedback@definitelymaybeagile.com, or come visit us at our website at definitelymaybeagile.com.

Dave (30:30): Thank you everybody. Thank you again, Derek. Thanks as always, Peter.

Derek (30:33): Yeah, thank you both, Peter, Dave, thank you for having me here.

Peter (30:36): You've been listening to Definitely Maybe Agile, the podcast where your hosts, Peter Maddison and David Sharrock, focus on the art and science of digital, agile, and DevOps at scale.